Mambrito B, De Luca C J
Electroencephalogr Clin Neurophysiol. 1984 Aug;58(2):175-88. doi: 10.1016/0013-4694(84)90031-2.
In the present paper we have described a system for acquiring, processing and decomposing EMG signals for the purpose of extracting as many motor unit action potential trains as possible with the greatest level of accuracy. This system consists of 4 main sections. The first section consists of methodologies for signal acquisition and quality verification. Three channels of EMG signals are acquired using a quadripolar needle electrode designed to enhance discrimination among different MUAPs. An automated experiment control system is devised to free the experimenter from the burden of experiment detailed surveillance and bookkeeping; and to allow on-line assessment of the EMG signal quality in terms of decomposition suitability. The second section consists of methodologies for signal sampling and conditioning. The EMG signal is bandpass filtered (between 1 kHz and 10 kHz), sampled and compressed by eliminating parts of the signal under a preset threshold level. The third section consists of a signal decomposition technique where motor unit action potential trains are extracted from the EMG signal using a highly computer assisted interactive algorithm. The algorithm uses a continuously updated template matching routine and firing statistics to identify MUAPs in the EMG signal. The templates of the MUAPs are continuously updated to enable the algorithm to function even when the shape of a specific MUAP undergoes slow variations. The fourth section deals with ways in which to analyze and display the results. The more frequently used representation formats are: (1) display of MUAP wave shapes; (2) impulse trains representing motor unit firings; (3) IPI plots where time interval between successive firings of the same motor unit is plotted vs. time of the muscle contraction; (4) firing rate plots where the estimated time-varying mean firing rate of the detected motor units is plotted vs. time of the muscle contraction. The performance of the system has been tested in terms of: (1) consistency among results obtained by different operators; (2) accuracy evaluated on synthetic EMG signal; (3) indirect measure of accuracy on real EMG signal by comparing results pertaining the same motor unit action potential trains derived by two EMG signals, independently and simultaneously recorded from two different electrodes.
在本文中，我们描述了一种用于采集、处理和分解肌电信号的系统，目的是尽可能准确地提取出尽可能多的运动单位动作电位序列。该系统由4个主要部分组成。第一部分包括信号采集和质量验证方法。使用四极针电极采集三通道肌电信号，该电极旨在增强对不同运动单位动作电位的辨别能力。设计了一个自动实验控制系统，使实验者从详细的实验监测和记录工作中解脱出来；并允许根据分解适用性对肌电信号质量进行在线评估。第二部分包括信号采样和调节方法。肌电信号经过带通滤波（1 kHz至10 kHz之间），通过消除低于预设阈值水平的信号部分进行采样和压缩。第三部分包括一种信号分解技术，其中使用高度计算机辅助的交互式算法从肌电信号中提取运动单位动作电位序列。该算法使用不断更新的模板匹配程序和发放统计信息来识别肌电信号中的运动单位动作电位。运动单位动作电位的模板不断更新，以使算法即使在特定运动单位动作电位的形状发生缓慢变化时也能发挥作用。第四部分涉及分析和显示结果的方法。更常用的表示格式有：（1）运动单位动作电位波形显示；（2）表示运动单位发放的脉冲序列；（3）肌间潜伏期图，其中绘制同一运动单位连续发放之间的时间间隔与肌肉收缩时间的关系；（4）发放率图，其中绘制检测到的运动单位的估计时变平均发放率与肌肉收缩时间的关系。该系统的性能已在以下方面进行了测试：（1）不同操作者获得的结果之间的一致性；（2）对合成肌电信号评估的准确性；（3）通过比较从两个不同电极独立且同时记录的两个肌电信号得出的与同一运动单位动作电位序列相关的结果来间接测量真实肌电信号的准确性。
